Generated in the 'Between': The Journey of Psycholinguistics and the Possibility of Relational Dynamics
Language is the dwelling of Being (语言是存在之家); relationship is the field of freedom (关系是自由之场).
Author: Liangzhi (良之) Publication date: March 17, 2026
Introduction: A Discipline About "Language and Mind"
The fundamental question psycholinguistics seeks to answer is: How does language operate in the brain?
This question seems simple, yet it touches the core of human self-understanding. When we utter a sentence, what happens in the brain? When an infant learns its first word, what restructuring does its cognitive architecture undergo? When we hear the three words "I love you" (我爱你), is it the ears listening, or the heart?
Seventy years of psycholinguistic exploration have attempted to answer these questions using scientific methods. Its history is a chronicle of inquiry into "how humans become human."
This article will systematically trace the evolutionary脉络 of psycholinguistics from a cross-disciplinary perspective, and explore the theoretical resources and methodological insights it may offer to emerging interdisciplinary fields—especially "Relational Dynamics" (关系动力学), which the author is currently constructing.
This is destined to be a cross-boundary dialogue, but as psycholinguistics' own history has proven: the most profound insights are often born at the boundaries of disciplines.
I. Prehistory: Philosophical Roots and Empirical Germination (Ancient Era–1953)
1.1 Ancient Philosophers' Reflections on Language
Human思考 about language far predates the appearance of the discipline name "psycholinguistics."
Plato in the Cratylus (克拉底鲁篇) posed a question still debated today: Is the relationship between names and things "natural" or "conventional" (约定的)? If names are natural, then language should possess some universal structure transcending culture; if names are conventional, then language is merely arbitrary social custom.
The ghost of this question has wandered throughout psycholinguistics' history.
Aristotle went further. In On Interpretation (解释篇), he proposed:
"Spoken words are the symbols of mental experience, and written words are the symbols of spoken words."
This triadic structure of [Reality–Mind–Language] foreshadowed the core question of modern semantics: How do linguistic symbols connect the external world to the internal mind?
值得注意的是, these ancient philosophers' reflections already touched the problem of the "between" (之间)—language is not an isolated existence; it lives between speaker and listener, between mind and world.
1.2 The Four Major Intellectual Currents of the Nineteenth Century
The birth of modern psycholinguistics was the result of several scientific intellectual currents converging since the 19th century:
| Current | Core Contribution | Impact on Psycholinguistics |
|---|---|---|
| Comparative linguistics | Discovered regularities in language evolution | Suggested language rules relate to innate mental structures |
| Brain function localization | Broca's area (1861), Wernicke's area (1874) | First proved language is a biological brain function |
| Child language observation | Darwin's Biographical Sketch of an Infant (一个婴儿的传略) | Revealed staged and biologically driven language acquisition |
| Experimental psychology germination | Wundt proposed "sentences are differentiation of holistic mental representations" | Laid foundation for studying psychological processes of language production |
1.3 Why 1953 Is the "Year of Birth"?
Despite these discoveries, in the first half of the 20th century, a chasm separated linguistics and psychology.
- Structuralist linguistics (Bloomfield): refused to study unobservable "mental processes"
- Behaviorist psychology (Skinner): reduced language to chains of "stimulus–response"
The two disciplines lacked a common theoretical language and research program.
The term "Psycholinguistics" was first used by American psychologist Kantor in 1936 in An Objective Psychology of Grammar (语法的客观心理学). But as a marker of disciplinary birth, it was the "Psycholinguistics Workshop" held at Indiana University in the summer of 1953, and the 1954 volume Psycholinguistics: A Survey of Theory and Research Problems co-edited by Osgood and Sebeok on this basis.
Psycholinguistics was born. But at the moment of its birth, the coming revolution was already gestating.
II. First-Generation Psycholinguistics (1953–1957): Behaviorist Dominance
2.1 The Theoretical Triangle of the First Generation
The first generation's theoretical framework consisted of three pillars:
┌─────────────────────────────────┐
│ Behaviorist Psychology │
│ • Learning mechanism: imitation + reinforcement│
│ • Experimental paradigm: reaction time measurement│
└─────────────────────────────────┘
↓
┌─────────────────────────────────┐
│ Structuralist Linguistics │
│ • Language units: phonemes, morphemes, phrases│
│ • Analysis goal: observable formal structures│
└─────────────────────────────────┘
↓
┌─────────────────────────────────┐
│ Information Theory │
│ • Quantification unit: bit │
│ • Core concepts: channel capacity, redundancy│
└─────────────────────────────────┘
The deep logic of this triangular structure: Language can be decomposed into measurable units, and the transformations between these units can be formally described.
This was the first attempt at scientification, but it omitted the most crucial thing—meaning.
2.2 Osgood's "Mediation Process Theory"
Charles Osgood attempted to find a reasonable place for semantics without violating behaviorism's basic principles.
His "mediation process theory" (中介过程理论) proposed: a word symbol does not directly elicit external behavioral responses, but first triggers an internal "mediating response" (中介反应) within the individual—a truncated, internal psychological process. This mediating response then serves as a "mediating stimulus" (中介刺激), ultimately eliciting external behavior.
Stimulus → [Mediating Response / Mediating Stimulus] → External Behavior
↑
Psychological reality of semantics
Osgood's other contribution was the Semantic Differential Scale (语义差异量表): through three dimensions—"Evaluation" (good–bad), "Potency" (strong–weak), "Activity" (active–passive)—he converted subjective "sense of meaning" into quantifiable psychological spatial coordinates.
Plotting "mother" (母亲) and "father" (父亲) on a coordinate graph reveals their distance in semantic space—the beginning of semantic quantification research.
2.3 The Fatal Limitations of the First Generation
Although first-generation research advanced quantification, behaviorist models had致命缺陷:
❌ Cannot explain language's creativity (创造性) Skinner's reinforcement theory can explain parrot-like imitation, but cannot explain why a three-year-old can utter ["Mom, I dreamed of flying penguins yesterday"]—a sentence never heard before, yet composed from limited vocabulary into a novel expression.
❌ Cannot explain language's recursiveness (递归性) Chomsky demonstrated the problem with such a sentence: ["John says Mary thinks Peter believes Susan will come."] This infinitely嵌套 structure cannot be acquired through imitation.
The enormous theoretical tension finally catalyzed the most剧烈 paradigm revolution in psycholinguistics history.
III. Cognitive Revolution and the Second Generation (1957–1970s): Chomsky's Challenge
3.1 The 1957 Turning Point
In 1957, 25-year-old Chomsky published a booklet of less than 120 pages: Syntactic Structures (句法结构).
This publication was not only a turning point in linguistics history but also宣告了 the arrival of psycholinguistics' second generation. Chomsky's Transformational-Generative Grammar (转换生成语法) included core ideas:
| Concept | Content | Significance |
|---|---|---|
| Competence vs. Performance (语言能力 vs. 语言表现) | Implicit knowledge vs. actual use | Research focus shifted from behavior to mind |
| Deep structure vs. Surface structure (深层结构 vs. 表层结构) | Meaning determined by deep, form by surface | Explains generation mechanism of synonymous sentences |
| Universal Grammar (UG) (普遍语法) | All human languages share a set of abstract principles | Provides theoretical basis for "Language Acquisition Device" |
3.2 The "Fatal Blow" to Behaviorism
In 1959, Chomsky published a lengthy review of Skinner's Verbal Behavior (言语行为). This review is regarded as one of the most important documents in cognitive science history.
Chomsky's core argument was the "Poverty of the Stimulus" (刺激贫乏论):
The language input children接触 is limited, error-filled, and fragmented. Through imitation and reinforcement alone, it is absolutely impossible to acquire a complex grammar system that can generate infinite sentences.
More致命, children can correctly understand and use sentences they've never heard—such as "Who stole whose what?" type嵌套 structures.
The only explanation: children's brains must possess some innate, language-specific acquisition device. Language is not taught—it "grows" (生长).
This review was a devastating blow to behaviorism. Skinner's theory从此 exited the mainstream of language research; cognitive psychology formally entered the historical stage.
3.3 The Rise and Fall of the DTC Hypothesis: A Classic Case of "Conjecture and Refutation"
Chomsky's theory sparked psycholinguistics' most famous experimental hypothesis—the "Derivational Theory of Complexity" (DTC) (衍生论复杂度假设).
The hypothesis claimed: a sentence's psychological processing difficulty is proportional to the number of transformation steps it undergoes in generative grammar.
Early reaction-time experiments似乎 supported DTC. But subsequently, refuting evidence accumulated:
- When a sentence's semantic relations are "irreversible" (e.g., ["The apple was eaten by the little boy"]), passive-sentence processing time was not significantly longer than active-sentence processing
- Contextual cues, semantic plausibility, and world knowledge influence can easily dissolve processing load from syntactic transformations
The refutation of DTC was a watershed in psycholinguistics history. It marked the discipline beginning to摆脱 its依附地位 of "doing experimental verification for linguistic theory," moving toward the third generation with independent explanatory models.
Rules linguists write on paper are not the same rules the brain runs.
3.4 The 1975 Royumont Debate: The Century-Defining Clash of Innate vs. Acquired
In 1975, at the Royumont Abbey (罗约蒙修道院) in France, a跨-century debate occurred:
| Opposing sides | Core stance |
|---|---|
| Piaget | Cognitive development is gradually constructed through children's interaction with the environment; language ability is a natural extension of general cognitive ability |
| Chomsky | Language ability is a special, modular, innate cognitive capacity |
The debate lasted three days, touching epistemology, biology, linguistics, and psychology. Piaget's constructivism and Chomsky's innatism激烈 clashed, yet neither could说服 the other.
Jerry Fodor played an important role in this debate. His "Modularity of Mind" (心智模块性) hypothesis argued: the mind is not a unified blank slate, but composed of a series of functionally specialized "modules." The language module is the most典型 example—fast, automatic, and informationally encapsulated from other cognitive systems.
The Royumont debate produced no victor. But it profoundly influenced subsequent cognitive science development.
Today's mainstream view: there are both innate language tendencies and acquired learning mechanisms—the two are not mutually exclusive but complementary and协同.
IV. Independent Development and Diversification (1970s–1990s): Third-Generation Psycholinguistics
Entering the 1970s, psycholinguistics formally became an independent empirical discipline with its own research paradigms, core questions, and explanatory frameworks.
4.1 Language Comprehension Research: From "Syntax-First" to "Interactive"
Garden-path sentence research (花园小径句研究) revealed a关键 finding:
[The horse raced past the barn fell.]
When readers first encounter this sentence, they likely建立 a "horse running" interpretation frame after "raced," then discover something is off upon reaching "fell"—"raced past the barn" is actually a relative clause modifying "horse," and the sentence's main verb is "fell."
Readers must backtrack, reanalyze, and修正 their initial understanding. This garden-path sentence research revealed the dynamic process of sentence comprehension:
The brain doesn't wait until the sentence ends to begin analysis;
it listens and guesses, parsing in real time.
This "immediacy" (即时性) is a fundamental feature of language comprehension and a核心 discovery of psycholinguistics.
4.2 Levelt's "From Intention to Articulation" Model
William Levelt's 1989 Speaking: From Intention to Articulation (说话：从意念到话语) was a milestone in language production research. His four-stage model remains the gold standard in the field:
Conceptualizer → Formulator → Articulator → Articulator
↓ ↓ ↓ ↓
Preverbal message Surface structure Phonetic plan External speech
(~hundreds of ms) (200-300ms) (100-200ms) (real-time)
The model's elegance lies in its explanation of the "Tip-of-the-Tongue" (TOT) phenomenon (话在嘴边现象):
TOT state is a "断点" in language production: the person clearly knows "what word I want to say," knows its meaning, knows its length, even knows its first letter, yet cannot produce its complete pronunciation.
Levelt's theory perfectly explains this: lexical retrieval分为 two stages—first retrieving the "lemma" (词条) (containing semantic and syntactic information), then retrieving the "lexeme" (词形) (containing pronunciation information). TOT state is the断点 where the first stage succeeds but the second fails.
4.3 The ERP Revolution: Seeing the Brain's "Language Moments"
In 1980, Marta Kutas and Hillyard, while conducting a language experiment, discovered an unexpected脑电波 change:
When a semantically inconsistent word appeared in a sentence (e.g., "I drink coffee with salt" [我喝咖啡加盐]), approximately 400 milliseconds later, a明显的 negative deflection appeared.
She named this component N400 (N for Negative, 400 for approximately 400ms latency).
N400's discovery opened a window for real-time observation of brain language processing. Subsequently, researchers discovered:
| Component | Latency | Sensitive to | Functional significance |
|---|---|---|---|
| ELAN | ~150ms | Early syntactic violation | Automatic process of initial syntactic analysis |
| N400 | ~400ms | Semantic inconsistency | Difficulty of lexical-semantic integration |
| P600 | ~600ms | Syntactic complexity/violation | Reanalysis or repair of syntactic structure |
These ERP components form a精细 [Electrophysiological Map of Language Processing] (语言加工电生理图谱).
ERP technology's revolutionary nature lies in its temporal resolution—millisecond-level precision enables tracking the real-time process of language processing. A sentence's processing can complete within hundreds of milliseconds; only ERP-like techniques can capture the dynamic细节 of this process.
V. Contemporary Psycholinguistics (2000s–Present): Methodological Revolution and Cross-Disciplinary Integration
5.1 Revolutionary Breakthroughs in Research Methods
Contemporary psycholinguistics laboratories have become high-tech集成地:
| Technology | Core advantage | Typical application |
|---|---|---|
| Eye tracking | High spatial resolution, good ecological validity | Cognitive resource allocation during reading |
| fMRI | High spatial resolution (millimeter-level) | Brain region localization of language networks |
| MEG | Balances temporal + spatial resolution | "When" and "where" of language processing |
| Hyperscanning (超扫描) | Simultaneously records multiple brains | "Inter-brain synchronization" (脑间同步) during interaction |
Hyperscanning technology is the most important technological breakthrough of the last decade, because it extends research from "single brains" to "between brains"—this is precisely the starting point for "relational" research.
5.2 Statistical Learning: Bridging "Innate–Acquired"
"Statistical learning" (统计学习) has become one of the most核心 research热点 in contemporary psycholinguistics.
In 1996, Jenny Saffran and colleagues conducted a classic experiment: letting 8-month-old infants listen to a continuous artificial speech stream with no pauses or intonation cues—the only线索 was "transition probabilities" (转换概率) between syllables.
For example:
• "pa-bi" always appears together (transition probability = 1.0)
• "bi-ku" also always appears together
• But after "ku," either "da" or "go" may follow (transition probability < 1.0)
After two minutes of listening, infants preferred listening to "words" (pabiku) like high-transition-probability syllable sequences rather than "non-words" (bikuda) like low-transition-probability sequences.
This experiment震惊了 academia: 8-month-old infants can, in just two minutes, relying solely on statistical regularities, "segment" words from continuous speech streams.
This means statistical learning is a fundamental capacity for human language acquisition. Subsequent research found:
- Individual differences in statistical learning ability affect language learning outcomes
- Statistical learning may not be a single "general learning mechanism" but composed of multiple相对独立 subsystems
- Statistical learning ability can predict second-language grammar learning outcomes
These findings have深远意义 for understanding the "innate–acquired" question:
Chomsky says "poverty of the stimulus," so innate grammar must exist. But statistical learning research shows children can extract far richer statistical information from input than we imagined. 所谓 "innate language ability" may be high sensitivity to statistical regularities—this sensitivity may be innate, but what it learns is acquired statistical structure.
5.3 The Impact and Opportunities of Large Language Models
The emergence of large language models (LLMs) represented by GPT is both冲击 and机遇 for psycholinguistics.
Impact: GPT-4 can write poetry, translate, program, and pass the bar exam. In grammar tests, it surpasses most humans. This has sparked激烈 debate: Do LLMs truly "understand" language? Or are they just "stochastic parrots" (随机鹦鹉)?
Opportunity: Computational psycholinguists discovered that the "Surprisal" (惊喜值) LLMs compute for the next word in a corpus correlates significantly positively with the N400 amplitude produced by human brains.
A word that is more "unexpected" in context:
• Humans: larger N400 amplitude
• LLMs: higher computed Surprisal
This means: LLMs' prediction mechanism may, to some extent, simulate the human brain's "prediction engine."
But despite LLMs demonstrating惊人 capabilities at grammatical and semantic levels, there are根本差异 between them and humans:
| Dimension | Humans | LLMs |
|---|---|---|
| Embodiment (具身性) | Has body, senses, emotional experience | No body, only processes symbols |
| Sociality (社会性) | Has childhood, attachment, intimate relationships | No social experience, no relational history |
| Intentionality (意向性) | Speaks with intention, beliefs, desires | Generates text without "wanting to tell you" intention |
These differences迫使 psycholinguists to reflect more deeply:
Which parts of human language comprehension must necessarily depend on body, context, and social interaction?
VI. Insights from Psycholinguistics for Relational Dynamics
6.1 The Four-Step Path from "Human Inquiry" to "Legitimate Discipline"
Psycholinguistics' history can be概括 as four stages:
┌─────────────────────────────────────┐
│ Step 1: Establish an axiomatic meta-language│
│ • Chomsky: formal tools of generative grammar│
│ • Provide precise, operational conceptual framework for language research│
└─────────────────────────────────────┘
↓
┌─────────────────────────────────────┐
│ Step 2: Introduce empirical methods│
│ • Miller: information theory + reaction time measurement│
│ • Eye tracking, ERP, fMRI sequentially introduced│
│ • Push from philosophical speculation to precise measurement│
└─────────────────────────────────────┘
↓
┌─────────────────────────────────────┐
│ Step 3: Find psychophysiological correspondences│
│ • Broca's area, Wernicke's area│
│ • N400, P600, ELAN│
│ • Convert "language" into locatable brain structures│
└─────────────────────────────────────┘
↓
┌─────────────────────────────────────┐
│ Step 4: Disciplinary institutionalization│
│ • Labs, journals, societies, doctoral programs│
│ • Transform personal thought into public knowledge│
└─────────────────────────────────────┘
These four steps are an极其珍贵 template for anyone attempting to establish a new discipline.
6.2 What Can Relational Dynamics Learn?
"Relational Dynamics" (关系动力学) seeks to answer a根本 question: How does the "between" (之间) become real (实在)?
Can that看不见 yet可感 [relational field] (关系场) be scientifically studied?
From psycholinguistics' history, a feasible path can be seen:
Step 1: Establish a Meta-Language
We need to find a set of precise, operational formal tools for "relationship"—just as generative grammar provided for language. Category theory, graph theory, or dynamical systems theory may be suitable mathematical母语.
Step 2: Introduce Dynamical Equations
Relationships are not静态 structures but动态 processes. We need mathematics to describe their evolution—generation, maintenance, transformation, dissolution. Decay functions, attractor models, phase-transition theory etc. are borrowable tools.
Step 3: Find Psychophysiological Correspondences (Most Crucial)
If the "relational field" is real (实在), it must have manifestations in brain and body:
- When a relationship is harmonious, do two people's heart rates synchronize?
- When a relationship is tense, is their skin conductance abnormal?
- When "reciprocal imbalance" (互惠失衡) occurs, does the brain produce fluctuations类似 to N400 or P600?
Statistical learning research provides an important启发: Can statistical learning also extend to "relational statistical learning" (关系统计学习)?
Are we unconsciously learning "the grammar of others' desires" (他者欲望的语法)—when approaching will be accepted? When expression will be heard? When silence is safe? When retreat is needed?
If the answer is affirmative, then psycholinguistics' experimental paradigms—artificial grammar learning, transition probability measurement, ERP recording—can be directly applied to relational research.
Step 4: Disciplinary Institutionalization
This step is还很遥远. But if the first three steps succeed, if there truly is evidence that the "relational field" is real, measurable, and has psychophysiological foundations, then establishing a "Relational Dynamics" discipline is no longer a castle in the air (空中楼阁).
6.3 A Concrete Research Proposal
Based on the above reflections, a concrete research proposal can be提出:
[Research Question]
Do anxiously attached individuals and securely attached individuals differ in "relational statistical learning" ability?
[Core Hypothesis]
Anxiously attached individuals have超常 sensitivity to relational signals—especially signals of "possible rejection."
This sensitivity may originate from their different way of learning relational statistical regularities:
• They learn "danger signals" faster
• But learn "safety signals" more slowly
[Experimental Design]
1. Task: Social version of artificial grammar learning
• Screen sequentially presents interaction symbols: "approach," "smile," "withdraw," "frown"
• Transition probabilities between symbols simulate real relational statistical structure
2. Measurement:
• Behavioral: speed of pattern extraction, prediction accuracy
• Physiological: synchronously record heart rate variability, skin conductance response
• Neural: record ERP, observe whether there is a "relational prediction error" component类似 to N400
[Expected Results]
• Anxiously attached individuals learn "rejection"-related sequences faster, learn "acceptance"-related sequences slower
• Their "relational N400" is more sensitive to negative prediction errors
[Theoretical Significance]
If this research succeeds, it would be the first batch of empirical evidence for "Relational Dynamics"—
proving the [relational field] is not only可感 but measurable.
Conclusion: Language, Relationship, and Freedom
The core insight of seventy years of psycholinguistic exploration:
Language is not an island; it flows between persons, generated between "I" and "You."
Heidegger said: "Language is the dwelling of Being" (语言是存在之家). Language lets us exist because language lets the world disclose itself to us. But language's meaning is not in dictionaries—it is in dialogue; not in single minds, but in the interpersonal "between" (之间).
Chomsky in his later years repeatedly强调: The essence of language is freedom (自由).
Animal communication systems are closed and fixed; human language is open and creative. We can utter sentences never heard, understand expressions never seen—this "creativity" is not conditioned reflex but the biological foundation of freedom.
The essence of education is likewise freedom—not灌输 dogma, but helping people discover truth and exercise freedom.
This proposition deeply resonates with psycholinguistics' core inquiry:
How do we create infinite meaning with finite rules? How do we connect each other's minds with sounds and symbols?
If language is the biological foundation of freedom, then "relationship" may be the social space of freedom.
In relationships, we are both shaped and can transcend. Good relationships—whether academic collaboration or daily interaction—should make each other freer, not more bound.
衡门之下，可以栖迟；(Under a humble gate, one may dwell;) 舞雩之上，可以咏归。(Upon the rain-dance terrace, one may sing and return.)
Language is the dwelling of Being; relationship is the field of freedom. (语言是存在之家，关系是自由之场)
May we, in the "between" (之间), generate more integral selves and freer connections.
Selected References
- Chomsky, N. (1957). Syntactic structures. Mouton.
- Chomsky, N. (1959). Review of Skinner's Verbal behavior. Language, 35(1), 26-58.
- Kutas, M., & Hillyard, S. A. (1980). Reading senseless sentences: Brain potentials reflect semantic incongruity. Biological Psychology, 11(2), 99-116.
- Kutas, M., & Federmeier, K. D. (2011). Thirty years and counting: Finding meaning in the N400 component of the event-related brain potential (ERP). Annual Review of Psychology, 62, 621-647.
- Levelt, W. J. M. (1989). Speaking: From intention to articulation. MIT Press.
- Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274(5294), 1926-1928.
- Gui Shichun (桂诗春). (2000). New Psycholinguistics (新编心理语言学). Shanghai Foreign Language Education Press.
Copyright Notice: This is a preview translation — Chinese original is the authoritative version. Copyright belongs to Guangzhou Phaenarete AI Technology Co., Ltd. Unauthorized reproduction, citation, or distribution is prohibited.